Reflections on a Puritan New Year
New Year’s is a natural time to reflect on the past twelve months and what we can expect in the twelve to come. In retrospect, there were three controversial bets that were proven out in 2012, bets that will continue to shape the storage market for the next several years:
- Flash memory demands an architectural rethink − Storage software designed for mechanical disk will not make the leap to the solid-state future. Witness the two market leaders in disk-centric storage software spending their treasure and talent on a clean slate approach for flash: EMC acquiring XtremIO and NetApp’s Project Mars. All major storage vendors have realized or will soon realize that all-flash arrays must be recrafted from the ground up. Pure has done it. EMC and NetApp are doing it. The remaining disk-centric holdouts are at risk of missing the boat.
- Data reduction is mandatory for primary solid-state storage − Very fast (submillisecond) inline deduplication and compression are essential to (following the Pure motto) deliver “flash for the cost of disk.” Based on 100s of customer and partner deployments, the database and virtualized workloads Pure targets support sufficient data reduction (4X or greater), that we really can offer all-flash storage below the price of performance disk. Hence the “no brainer” elevator pitch—10X faster, 10X more power & space efficiency, 10X simpler, more reliable, and all at comparable cost.
In 2013, EMC is expected to launch a product based on the XtremIO acquisition that will duplicate some portion of Pure’s inline data reduction functionality launched in 2011. Data reduction is also apparently enshrined within NetApp’s Project Mars. Now every flash storage vendor is working their tails off to attempt to follow suit. The challenge is that it took Pure and XtremIO years to get our algorithms right, and thus the race is on to see who else has the software chops to get into the game.
- For storage, software still trumps hardware – We at Pure have not been surprised that with the right software, commodity SSDs, commodity interconnects (Infiniband & SAS), and Intel processors, the FlashArray is performance competitive with far more expensive proprietary hardware designs, and even with server-local PCIe flash cards. While Pure’s marketing claims have been in the 100,000s of IOPS rather than the 1,000,000+ advertised by some of our competitors, we have consistently either won or been neck and neck in competitive benchmarks, and been well ahead on price performance.
For 2013, here are three new bets for your consideration:
- The winning recipe for all-flash storage is made clear − Vendors that are unable to deliver true enterprise high availability (HA), RAID6-style dual parity protection, and inline data reduction will start losing ground. HA and RAID6 reduce flash costs and increase reliability versus the naive approach of replicating boxes and basic mirroring. Mirroring requires 60% more flash but is less reliable in that there is no protection from overlapping failures. And flash-optimized inline data reduction accelerates performance by avoiding relatively expensive writes, and also saves wear and tear on the flash.
- The luster wears off of auto-tiering within a storage array − The problem with tiering within a volume is simply one of expectations: when your application expects disk-performance and instead have its request serviced from flash cache, you are thrilled. But as soon as you expect flash performance, a two orders of magnitude latency spike for missing cache and hitting disk becomes unacceptable. Such disparities in performance doomed disk/tape hybrid appliances—once a $1B+ business, because having tape in the latency path became unacceptable. The same thing will happen even more quickly for performance-intensive primary storage, particularly with flash now competitive with disk in price (thanks to data reduction algorithms that are incompatible with mechanical disk).
- VDI takes off with all-flash storage and inline data reduction − Virtual Desktop Infrastructure (VDI) got to critical mass in 2012 thanks to the appeal of Bring Your Own Device (BYOD) and the realization that VDI can actually improve the user experience on legacy PCs. Indeed all flash storage with inline deduplication has proven to be perfect for VDI—far lower cost than disk (due to high data reduction >10:1) and faster performance than a laptop’s local SSD! (Not so surprising when you consider a network round trip is a fraction the cost of reading or writing the flash, and so if you can do a better job managing the flash on the other side of the network, it’s a performance win for the client.) For more on this, please check out our newly-released all-flash reference architectures for VDI.
2012 has truly been a fabulous year for Team Puritan. Highlights include:
- Shipping our first generally available release (after two years of private betas), as well as launching HA, iSCSI/Ethernet support, and hyper-efficient snapshots (near zero space and performance overhead), arguably, one of the best years ever for a storage engineering team;
- Winning the Wall Street Journal’s Innovation Award, not just over the cadre of other storage start-ups, but also beating out dozens of highly innovative tech companies across the board—a gratifying result for our humble venture; and
- Poking fun at ourselves and the rest of the storage industry with our award-winning “You Know You Want Flash, Right?” viral video.
And it all came together with fantastic market reception for the Pure Storage FlashArray: Our customers and partners have enabled Pure Storage to arguably have the most successful opening night in storage history. While in fairness we are unable to make comparisons with other private companies, we know that we’ve made a bigger splash (in terms of inflation adjusted revenues) than did NetApp or Data Domain over their first few quarters, generally considered the fastest growers in storage history. Yes, it is far too early to be drawing meaningful comparisons—if we can make similar claims two years from now, that will be an impressive accomplishment indeed.
But, knocking on wood, it is hard to see how things could be going much better for Team Puritan just now. For such an unprecedented year, many thanks are owed:
- Thanks to our customers and partners, who made early bets on an audacious vision—that solid-state flash would supplant mechanical disk in primary storage;
- Thanks to the Herculean efforts of all of the Pure employees—we are thrilled and gratified to have built such a deeply talented team, wholly committed to customer and partner success; and
- Thanks to our investors, who made putting this team together on this mission possible, and who continue to help keep our venture on the right path.
While we are extremely grateful for the success to date, we remain focused on the work ahead. No doubt we have a long, long way to go if we are to succeed in our mission of leading the industry transition from mechanical to solid-state storage. In 2013 expect to see Pure Storage
- Recruit 100 new employees to join our team;
- Plant our flag in Europe and Asia;
- Double our downtown Mountain View, California headquarters office space;
- Substantially boost the performance and scale of our technology; and
- Compete ever more aggressively with the incumbent vendors trying to squeeze a few more years of life out of their decade old mechanical storage legacies.
Disk is the new tape, and flash is the new disk. The future of storage is here today with Pure. With all-flash at the price of disk, what are you waiting for? Happy new year from all of us at Pure!